FI118242B - Management of speech frames in a radio system - Google Patents

Description

118242

Working with a speech frame in a radio system

Area

The invention relates to a method for processing a speech frame in a radio system, a radio system, a subscriber terminal of a radio system, and a network of a radio system.

Background

The speech to be transmitted by the radio station system is encoded in the transmitter by a speech encoder. An example of a speech encoder is the Terelestial Trunked Radio (TETRA) ACELP (Algebraic Code Excited Linear Predictive) encoder.

10 The TETRA system is described in John Dunlop, Demessie Girma, James Irvine, Digital Mobile Communications and the TETRA System, John Wiley &

Sons Limited 1999, ISBN 0-471-98792-1, which is incorporated herein by reference. The receiver has a speech decoder which converts received speech descriptive information back into human-understood speech. If the speech encoder 15 also includes a speech decoder, the device is called a speech codec.

The digital information representing the speech is coded in the channel to prevent the harmful effects of noise in the radio path. The channel codes may be error detection and / or error correction. The channel codes can in principle be divided into block codes and convolutional code. 20 dei. Both encoding methods can also be used simultaneously. One block code "V" is cyclic redundancy check (CRC).

In addition, convolutional coding and its various modifications, such as punctured convolutional coding, are typically used. A typical convolutional *. * ·: Code rate, i.e. the ratio of the number of user data bits • · · 25 to the coded data bits of a channel, is, for example, 1/2 or 1/3. · · [[[: Encoding is often used interleaving. In interleaving, consecutive bits are mixed for a longer period of time so that the current \ f # time fade on the radio path is not sufficient to make the radio signal for that period. ·· *. • • · [30 band decoding can still be eliminated.

: In addition to channel coding, the information can be encrypted to prevent eavesdropping. Encryption is generally accomplished by generating an encryption mask using the encryption algorithm using a specific encryption mask: which is associated with the X · 35 transfer information by an XOR operation (logical exclusive OR operation). The encryption can be done either before or after the channel .., .- 2 118242 coding. There may also be several levels of encryption, for example, encryption over a radio interface, and point-to-point encryption, for example, between two subscriber terminals. The receiver can be decrypted using the same encryption algorithm used in the transmission as long as the input parameters are the same.

An error-detecting channel code, such as a cyclic redundancy check calculation, generally detects bit errors in the channel-coded speech frame on the radio path, sending a Bad Frame Indication (BFI) to the speech decoder, telling the decoder not to attempt to decode the speech frame. In this case, the speech decoder generally uses a procedure in which the parameters of the previous intact frame are used to reduce gain. f

However, there are situations where, when bits of the speech frame are appropriately turned on the radio path, the speech frame no longer contains valid data that could be decoded by the speech decoder, even though the speech frame is decoded based on the channel code. Because of the very large number of possible bit combinations contained in the speech frame, for example, TETRA 2137-1 different combinations, it is virtually impossible to test all of them. In this case, it is possible that when attempting to decode a faulty speech frame, the output of the pu-20 decoder may become a very strong false signal, causing an unpleasant audio shock to the listener's ear, and at worst:: *: hearing loss.

* · * ·: The problem is exacerbated by the fact that, for example, in TETRA * * ·. ···. utilizing, at the request of customers, the maximum allowed power of the subscriber terminal speaker to maximize speech output in noise / s environments. Thus, the power limitation of the loudspeaker circuit itself, for example 100 dB or even 110 dB, does not solve the problem. The problem is also exacerbated by the fact that in many radio ice systems, when a speech frame is defective, the decoding of the previous speech frame is repeated with a speech decoder, so if * *: ** · 30 was defective, but based on the channel code, then two defective speech frame.

:! ·. Although channel coding would be able to detect errors, the second problem described below can cause audio shock. Usually in radio systems, such as the Global System for Mobile Corn (GSM) system, the encryption is decrypted before channel decoding. However, in TETRA, for example, decryption is only performed on channel decoding 118242 3. This may lead to a situation where the speech-containing speech frame is actually error-free based on the channel code as well, but for some reason, for example due to incorrect input parameters, the contents of the speech frame are completely blurred. If the speech decoder detects bad pu-5 frames based on the channel code, for example, by obtaining a bad frame indication from the channel decoder if necessary, it will not be able to recognize a decoded speech frame that does not receive a bad frame indication. In this case, the speech decoder attempts to decode the confused speech frame, which again results in audio shock. The problem is made more likely by the two-level browsing capability in TETRA: radio-boundary surface encryption and point-to-point encryption. In TETRA terminals, the user of the device may place his own encryption module for point-to-point encryption, whereby the device manufacturer has no way of verifying and testing the operation of the encryption module in question, since the encryption module already processes channel-decoded information. The encryption module is a black box for the device manufacturer, which ideally emits a speech-containing speech frame that has been decrypted. In addition, when updating the encryption keys, for example, errors may occur where the wrong encryption key is used as the input parameter of the encryption algorithm.

BRIEF DESCRIPTION: It is an object of the invention to provide an improved method, an improved · · · ·; radio system, improved radio system subscriber terminal, and improved * · *, · * ·. radio system network. One aspect of the invention is provided by the method of claim 1. One aspect of the invention is provided by a radio system according to claim 11. One aspect of the invention is a brochure according to claim 21.

One aspect of the invention is a radio system network according to claim 22. Other preferred embodiments of the invention are the subject of these claims.

30 The invention is based on the fact that, even if the speech frame is a channel: *. based on the encoding, the parameter descriptive of at least one speech in the channel decoded speech frame is examined, and a decision is made on the basis of T to determine whether it is worthwhile to attempt to decode the speech frame using a speech decoder. Audio shock is thus avoided by blindly relying on the ability of channel -: / ·: 35 encoding to detect errors. On the other hand, if the system was decrypted after channel decoding, then studying the parameters contained in the channel decoded and decrypted speech frame would also prevent a possible audio shock in the case where the channel decoded speech frame was error-free but decrypting the speech frame.

The most significant advantage of the invention is that, when used, it is possible to avoid audio shocks in the problem situations described above.

List of figures

Preferred embodiments of the invention will be described by way of example with reference to the accompanying drawings, in which: Figure 1 shows an example of the structure of a radio system; Fig. 2 shows a structure of a radio system network and a radio system subscriber terminal; Figure 3 illustrates the structure of a radio transmitter and a radio receiver; Figure 4 shows the channel coding of speech bits; FIG. 5 is a flowchart illustrating a method for processing a speech frame 15 in a radio system.

Description of Embodiments

Referring to Figure 1, the structure of a radio system will be described. The system to be described is TETRA, but it is to be understood that the embodiments are not limited to that system, but can be used in any radio system that exhibits at least one of the initial problems described, i.e., channel decoding non-ideality and / or decryption. · * ;;; ' The TETRA standard describes the six system components and their '··· * interfaces. The internal interfaces of the system components are not specified in the standard, so that manufacturers can implement the internal structure of each system component in the most efficient way. The system components are: '· * γ network 100, line station 108, mobile station 112, direct mode mobile station 114, * * .. .. 116, gateway 104, and network management unit. ···. ment unit 106.

* «'30 Network 100 is a TETRA network system comprising individual network elements, such as a support for implementing a radio connection illustrated in Figure 2.

»M

the station 212 with its antennas 210 and the mobile switching center 214. The network 100 corresponds approximately to the base station. ·.: subsystem of the GSM system combined with the network subsystem. In Fig. • · · 35 1, this combination is illustrated by reference numeral 102. In addition to reference numeral 102, 118242 5, the network 100 also includes a network management unit 106 and a gateway 104.

As shown in Figure 1, the TETRA network 100 may also be connected to another TETRA network 120, wherein the second TETRA network 120 has a radio connection 122 to a TETRA subscriber terminal 124.

The control station 108 is in practice a terminal or dispatcher unit located in a control room controlling the TETRA network 100.

Gateway 104 enables calls between users of TETRA network 100 and non-TETRA users. Defining gateway 104 is necessary because other networks connected to the TETRA network 100 use incompatible information formats and communication protocols, which requires translations or conversions. Figure 1 illustrates, as an example of a non-TETRA network, a public telephone network 130 in which a normal telephone 132 can be accessed from a TETRA subscriber terminal 112.

Other non-TETRA networks that can be accessed through gateway 104 are ISDN (Integrated Services Digital Network) and Public Data Network.

Network management unit 106 is used to execute the network administrator locally and remotely. The network administrator includes, for example, monitoring of error situations 20, system configuration, billing, performance measurement, and planning.

A: The subscriber terminal 112 provides a radio connection 110 to a network 100.

* * * *. The subscriber terminal 112 may be portable or in-vehicle. Suoramoo-, ···. the subscriber terminal 114 may be in radio communication 118 directly with the other • · 25 direct mode subscriber terminal 116 without the network 100 being utilized in any way to implement the radio connection 118. The radio connection 118 may be a point-to-point connection or a multipoint connection. Also, a device which combines both a normal subscriber terminal 112 and a direct mode subscriber terminal 114 is possible. A normal radio connection 110 using network 100 ·: * ·· 30 uses the trunking mode air interface, and the radio link 118 between the two direct mode subscriber terminals 114, 116 uses the direct mode air interface.

:! ·. The connection * · · between two direct mode subscriber terminals 140, 148 may also be implemented using an independent radio repeater 144. In this case, the radio connection 142 of the first ** subscriber terminal 140 and the radio repeater 144 *: * ': 35 utilizes a modified direct mode air interface, as well as the radio connection 146 of the second subscriber terminal 148 and the radio repeater 144. 152 operation. The direct mode subscriber terminal 156 then uses a modified direct mode air interface to establish a radio connection 154 with the gateway radio repeater 152. The radio connection 150 of the gateway radio repeater 152 is implemented with the network 100 using the trunking mode air interface.

A single call can have multiple encryptions using different encryption keys and encryption algorithms at the same time. For example, in Figure 1, the connection 110 between the subscriber terminal 112 and the network 100 has radio interface encryption. Then, the connection 100 between the network 100 and the gateway radio repeater 150 also has radio interface encryption. Also, the connection 154 between the gateway radio reproducer 152 and the direct mode subscriber terminal 156 has radio interface encryption. Additionally, point-to-point encryption may be provided between the subscriber terminal 112 and the direct mode subscriber terminal 156. That is, four different encryptions to make one call. The situation could also be worse 15, for example, from the gateway radio repeater 152 having a connection to the radio repeater 144, which would still be connected to the direct mode subscriber terminal 148. This would allow up to five different encryptions implemented in different calls with different encryption masks.

2, the subscriber terminal 112 has an antenna 200, a transceiver 202, and a control section 204. Other components of the subscriber terminal 112, such as means for implementing a user interface, are not described herein.

: Here, TETRA is not introduced anymore, but readers are advised to * ·· ·. please check out the book '' Digital Mobile Communica- 'mentioned above. ···. tions and the TETRA System. "

Referring now to Figure 3, the structure of a radio transmitter and a radio receiver is illustrated.

* · *! ”* The upper part of Figure 3 illustrates the structure of the transmitter for simplification. It is understood that the transmitter also includes other functions and components, but they are not relevant to this review. Speech 320 is encoded by a speech M Γ: * ·· 30 encoder, for example, the ACELP encoder mentioned above, which is described in the book Digital Digital Communications and the TETRA System.

; ! ·. The structure of the ACELP codec is described in the ETS 300 *: ETS 300 *: "* 395-1, May 1997," Terrestrial Trunked Radio (TETRA); Speech CODEC for 35 full -rate traffic channel; Part 1: General Description of Speech Functions. "

• · · «118242 7

Next, the bit stream produced by the speech codec, which describes speech in the form of various parameters, is passed to the encryption block 302, whereby the bit stream is encrypted as described at the beginning by attaching an encryption mask produced by the encryption algorithm. ; The encrypted bitstream is then channel coded in the channel encoder 304, for example in TETRA, first block coding in the form of CRC and then convolutional coding. This is illustrated in more detail in Figure 4, which shows the channel coding of speech bits. The ACELP codec produces 137 bits for each part of speech of 30 milliseconds corresponding to a bit rate of 4.567 kbit / s.

10 Bits twice 137 bits, since two speech frames produced by the ACELP codec are placed in one burst of the radio interface. The bits of a single speech frame are divided into three different classes based on the bit sensitivity. The 30 most sensitive bits are placed in category 2, 56 bits are placed in category 1, and the least error 51 bits are placed in class 0. Bits 15 of class 2 are added with an eight-bit cyclic redundancy check and four tail bits. Thus, there are 72 bits in class 2 that are subjected to 8/18 convolutional coding, i.e., each of the eight data bits corresponds to 18 channel encoded data bits, i.e. the final number of bits in class 2 is 162 bits. Class 1 bits are subjected to 2/3 convolutional encoding, i.e., 112 data bits become 168 channel 20 encoded data bits. Class 0 bits are not protected at all by channel coding, i.e. they enter a total of 102 bits per channel. The total number of bits is \: thus 162 + 168 + 102 = 432 bits divided by the normal uplink; V in a burst or downlink burst into two 216-bit fields. '"1 • · · III The 137 bits produced by the ACELP codec are described in the following table.

• I

25 bits 1. The bits consist of parameters that are classified as belonging to either the * 1 · 'lt [: filter or one of the four subframes. The filter parameters are * ··· 2 codebook indexes (Codebook Index) of ten LSP1- · · · LSP10 (LSP = Line Spectrum Pair). Each subframe has the following parameters: Pitch Delay, codebook indexes for 4 different 30 pulses, Pulse Global Sign, Pulse 3: Shift, and codebook index for Gains.

• ♦ 1 • · • · · * · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · to · · 8283 ·

Parameter Category Parameter Name__Bit number

Filter Codebook Index: LSP1-LSP3 8

Codebook Index: LSP4-LSP6 9 _Codebook Index: LSP7-LSP10 9 _

Subframe 1 Volume Delay 8

Codebook Index: Pulse 4 3

Codecode Index: Pulse 3 3

Codecode Index: Pulse 2 3

Codecode Index: Pulse 1 5

Global pulse sign 1

Pulse Shift 1 __ Codebook Index: Gain 6

Subframe 2 Volume Delay 5

Codebook Index: Pulse 4 3

Codecode Index: Pulse 3 3

Codecode Index: Pulse 2 3

Codecode Index: Pulse 1 5

Global pulse sign 1

Pulse Shift 1 __Codebook Index: Confirmations 6_

Subframe 3 Tone Delay 5 • Codebook Index: Pulse 4 3 »• I»

Codebook Index: Pulse 3 3 ···. Codebook Index: Pulse 2 3 ♦ ·

Codebook Index: Pulse 1 5 ·· «

Pulse Global Character 1 *: "* Pulse Shift 1 • * * ··· * Codebook Index: Confirmations 6 _

Subframe 4 Volume Delay 5 ··: *** '; Codebook Index: Pulse 4 3 A · · '

Codebook Index: Pulse 3 3: Codebook Index: Pulse 2 3] · * ·] Codebook Index: Pulse 15 ♦ · * "Global Pulse Character 1

Pulse Shift 1 • ·:. '* I _Codebook Index: Confirmations 6 _ 118242 9

Table 1: 137 bits produced by the ACELP codec

In speech decoding, the parameters described in Table 1 are decoded into data and the speech is reconstructed by a synthesis filter. The main class 2 bits are located in the following speech frame parameters: - codebook index: LSP1-LSP3, speech frame bits B1-B4; - codebook index: LSP4-LSP6, speech frame bits B9-B12; - codebook index: LSP7-LSP10, speech frame bits B18-B21; 10 - Volume delay for subframe 1, speech frame bits B27-B32; - In each subframe 1-4 codebook index: gain, speech frame bits B51-B53, B73-B80, B105-B107 and B132-B134.

Finally, the encrypted and channel coded frame is transmitted to the radio path 110 15 by modulating it in modulator 306. In TETRA, TDMA (Time Division Multiple Access) is used as a multiple access method, and π / 4-DQPSK (Differential Quaternary Phase Shift Keying) is used.

The bottom of Figure 3 illustrates the structure of the receiver in simplified form. It is understood that the receiver also includes other functions and structural components, but these are not relevant to this review. Referring also to the flow chart of Figure 5, which illustrates the flow; a system for processing a speech frame in a radio system.

The burst received from "V Radio path 110" is demodulated in demodulator I "" 308. Then, channel decoding is performed in channel decoder 310, i.e., cyclic redundancy checking is performed and the convolutional coding is decoded as a precursor using a Viterbi decoder. At this point, a method for actually processing a speech frame in a radio system is actually begun.

• · ·,

Execution of the method is started in block 500, and in block 502, a channel coded speech frame passed over the radio path to the channel decoder.

Next, the method has an optional decryption block 504, where the speech frame is decrypted, if one has been used.

***. *. Then, in block 506, it is checked whether the speech frame is error-free on the channel decoding basis. This can be accomplished, for example, by comparing * ··· * the calculated cyclic redundancy check with the 8: 8 cyclic redundancy check received from the channel to include ·: ··· 35 class 2 bits.

• · · • · 118242 ΊΟ> ··; 'ΐ

If the channel decoding, for example, a cyclic redundanssitar-inspection in a speech frame is defective, the block 506 goes as shown by arrow 520 to block in accordance with 514. The block 514 the erroneous speech frame does not attempt to encode the speech decoder 316 but: 5 A) The speech frame may be replaced with a previously received error-free speech with a speech frame, attenuating the gain.

B) Place a pause on the speech frame in question, that is, no sound is heard from the device speaker for 30 milliseconds.

C) The radio connection 110 can be disconnected if the radio connection is too poor, for example, if a number of faulty speech frames have already come in succession or over a certain period of time.

If the call terminates, the block 514 according to arrow 530 to block 516, where the embodiment of the method is terminated. If the call continues, 15 of the block 514 as indicated by arrow 528 to block 502 to process the next received speech frame.

If on the basis of the channel decoding the speech frame is correct, then we go from block 506 to block 522 as indicated by arrow 508. In block 508 is concluded from the value of the channel-speech frame of at least one polyurethane powder 20 contains a speech parameter frame to the speech decoder the speech decoder. For example, in TETRA, the speech:. ·, Descriptive parameter refers to the parameters produced by the ACELP-♦ ♦ ·. “V codec described in Table 1 above. This reasoning can be performed as described in U.S. Patent No. 6,021,385 (Järvinen et al.) By generating a probability distribution for the values of a pu * "*! 25 parameter, but with the difference that the method is not implemented by radio link quality. * * .. · but based on the success of channel decoding, except that the * * channel-protected error-sensitive bits contained in the speech frame are used in inference 508, i.e. TETRA also generates 30 kan 2 bits to aid in block 508. the difference with this US patent is also that the present US patent is unable to solve the problems presented in this application. US 6,021,385 is intended to improve the quality of speech decoding • · '··· * · information about the quality of the radio link, such as: ·: * ·: 35 receive power or bit error rate.

• · · * · * ·· '· 118242 11 if

In block 510, the result of the inference made in block 508 is checked.

If the inference, the speech frame does contain speech in the speech decoder 316, the routine shown by arrow 526 to block 512, where the decoded speech frame of the speech decoder 316. Then, if the call terminates, the block 512 is 5 as shown by arrow 532 to block 516, where the embodiment of the method is terminated. If the call continues, the block 512 according to arrow 534 to block 502 to process the next received Puheke-hystä. If the inference, the speech frame does not contain speech decoder the speech decoder 316, the routine 524 according to the arrow 10 on the Ku-vattuun to block 514, where the speech frame is not decoded but one of the replacement measures of block 514.

Thus, the radio system of Figure 3 further comprises means 314 for judging from the value of at least one speech descriptive parameter contained in the channel decoded speech frame whether the speech frame contains speech that can be decoded by the speech decoder 316 if the speech decoder 310 is error-free. Further, the speech decoder 316 is adapted to decode the speech frame if it is inferred that the speech frame contains speech that can be decoded by the speech decoder 316, and the speech decoder 316 is adapted not to decode the speech frame if the speech frame 206 does not contain speech. A potential decoder 312 is coupled to the channel decoder 310 and the inference means. ^ 314 in between, although in other systems other than TETRA the decryption device 312 * A may be located before the channel decoder 310. From the decryption device 312 to encryption • · · extracted speech frame is taken to the inferring means 314, and the direction of the arrow 324 mu kaisesti 25 to the speech decoder 316.

• ·

As indicated by arrow 322, the channel decoder 310 sends a bad Wednesday-frame in an indication to the speech decoder 316 if the base basis of the channel decoding the speech frame is defective. The speech decoder 316 will, in an optimal case, output the same-sounding speech 320 as originally transmitted at the transmitter.

* · #. ···. In a preferred embodiment, the inferring means 314 is adapted to transmit a bad frame assignment 324 to the speech decoder 316 if:.: I concludes that the speech frame does not contain speech that can be decoded by the speech decoder 316. a worthwhile attempt to decode a speech frame is one in which the inference ·, * means 314 is adapted to transmit a homing sequence 326 t · • · 118242 12 to the speech decoder 316 if it is deduced that the speech frame does not contain speech decodable by the speech decoder 316. is to put the speech decoder 316 in a known state, thereby enabling an output which does not cause an audio shock to be output to the loudspeaker. 5 In one preferred embodiment, the inference according to block 508 of the method is performed such that the inference means 314 This is accomplished, for example, by means of which the inference means 314 calculates the probability of the value of at least one speech-describing parameter in the inference. If the parameter describing the speech receives a highly unlikely value in that situation, it can be assumed that the speech frame does not contain speech. Of course, using two or more parameters for which probability is calculated will increase the certainty of the reasoning. The likelihood of a change in the value of a parameter describing a speech can also be calculated. Then, in the decision means 314, a threshold value is defined for the likelihood of changing the value of the 15 parameters to be compared over a given number of speech frames. The decision means 314 concludes that the speech frame does not contain speech that can be decoded by the speech decoder 316 if the change probability is less than a threshold value. By way of example, it is conceivable that during a call, the pitch of the speaker does not vary much. If the value of the pitch parameter changes too much between two consecutive speech frames, it can be concluded that:. has been corrupted either by radio path interference and / or decryption vir. because of their gentle actions. So it is not about the value of a parameter or • · · t '*' t that changes as such, but about the probability of a parameter value or * · Y \ 25 the probability of a parameter value changing.

*; ./ In principle, the probability calculation of the parameters is that · · there is always a correlation between the bits of the parameters. Probability calculation can be used to calculate probabilities between parameters within a single speech frame, or values between consecutive parameters of • • • *. · 30-frame speech frames. Of course, both methods of calculation can also be used simultaneously to confirm the result,. for example, by first calculating the probabilities between speech frames, and then, if the speech frame begins to appear corrupt, then calculating the probability of the · · **. ·· * second speech frame for at least one parameter. Comparing the values of consecutive *: * ·: 35-frame speech frames requires that the inference means have memory to store previous calculated values or whole speech frames.

The transmitter / receiver pair of Figure 3 may be, for example, two TETRA normal subscriber terminals 112, two TETRA direct mode states 5 and 114, a subscriber terminal 112 and a network 100, or a network 100 and a subscriber terminal 112. speech decoding may be performed in gateway 104 before going to non-TETRA network 130, or in some part of network 100, such as mobile switching center 214, before going to another TETRA network 120 or lead station 10,108.

The required functionality can be implemented, for example, as software executed in a general-purpose processor, whereby the required functionality is implemented as software components. A hardware implementation is also possible, for example as control logic built in ASIC (Application Specific Integrated Circuit) or differentiated components. Typically, all of the functions described in Figure 3 can be implemented as software. Thus, the inference means 314 relevant to the invention may also advantageously be implemented as software receiving input data from a decoder 312 and having an input interface 326 to a speech decoder 316. The software may be executed, for example, in the control section 204 of the subscriber terminal 112

Although the invention has been described above with reference to the example "V in the accompanying drawings, it is clear that the invention is not limited thereto but * 1 · V. '.' it can be modified in many ways by the * "'! 25 inventive ideas.

* · · ··· '• · · 1 · r · »· • · * · * 1 • ♦ · * · * · · · · · -'- f' • • 1 * · 1 • · • · · • · • · · · * ·.

• «• · * · · • ·.

• 1 • · · ·

Claims (22)

A method of processing a speech frame in a radio system, comprising: (502) a channel coded speech frame channel decoded over the radio path; characterized by: (506, 522) if the speech frame on the basis of the channel decoding is error-free, then (508) is determined on the basis of the value of at least one parameter describing tai and which is included in the channel decoded speech frame, whether the speech frame 10 contains tai which can be decoded with a speech decoder, and (510, 526) if the speech frame according to the decision contains tai which can be decoded with the speech decoder, say (512) the speech frame is decoded with the speech decoder, and (510, 524) if the speech frame does not contain tai which can be decoded with the speech decoder, say ( 514), the speech frame is not decoded in question, but none of the following is performed: a return sequence is transmitted to the speech decoder, a speech is inserted at the location of the speech frame in question, the radio connection is interrupted. Method according to claim 1, characterized in that the speech frame is encrypted, the encryption of the speech frame being canceled in the method (504). 3. A method according to claim 2, characterized in that (504) the speech frame encryption is canceled after the channel decoding (502) before the decision (508). • · · I. . *. Method according to claim 1, characterized in that: ···. the speech frame according to the decision (508) does not contain tai which can be decoded with the tai decoder, then an indication of a poor frame is sent to the speech decoder. 5. A method according to any of the preceding claims, characterized in that also symbols included in the speech frame and which are protected with channel coding are used in the decision (508). 6. A method according to any of the preceding claims, characterized in that the decision (508) is carried out by using the probability calculation. :! ·. 7. A method according to claim 6, characterized in that, at * f ·: the decision (508), the probability of the value of at least one parameter * * * is calculated describing tai. Method according to any of the preceding claims, characterized in that, in the decision (508), the probability of a change in the value of at least one parameter describing tai is calculated. Method according to claim 8, characterized in that a threshold value has been determined for the probability of a change in the value of the parameter below a certain number frame A. 4 5 10. A method according to claim 9, characterized in that if the probability of a change is less than the threshold value, it is concluded that the speech frame does not contain tai which can be decoded with the speech decoder. A radio system comprising a channel decoder (310) for decoding a channel coded speech frame over the radio path, and a speech decoder (316) for decoding the speech frame; characterized in that the radio system further comprises determining means (314) for deciding on the basis of the value of at least one parameter describing tai and which is included in the channel decoded speech frame, whether the speech frame contains tai which can be decoded by the speech decoder (316), in case the speech frame according to the channel decoder (310) is faultless; and the speech decoder (316) is arranged to decode the speech frame if the speech frame according to the decision contains tai which can be decoded by the speech decoder (316), and the speech decoder (316) is arranged not to decode the speech frame if the speech frame does not contain tai which can be decoded with the speech decoder * (316), wherein the decision means (314) are arranged to transmit a return sequence (326) to the speech decoder (316), the speech decoder (316) is arranged to insert into the speech a: Γ: pause at the location of the speech frame in question, or the decision means (314) is arranged to cause an interruption in the radio connection. ··· Radio system according to claim 11, characterized in that: ···! the speech frame is encrypted, the radio system comprising an encryption repeater • · s '.' (312) to cancel the speech frame encryption. Radio system according to claim 12, characterized in that the encryption canceller (312) is coupled between the channel decoder (310) and the decoding means (314). Radio system according to claim 12, characterized in that the J determining means (314) are arranged to transmit an indication (324) on a bad signal. frame to the speech decoder (316), if the speech frame according to the decision does not contain tai which can be decoded by the speech decoder (316). '* 35 Radio system according to any of the preceding claims 11-14, characterized in that the decisive means (314) also use symbols 118242 which are included in the speech frame and which are protected with channel coding in the decision. Radio system according to any of the preceding claims 11-15, characterized in that the decision means (314) perform the decision by using probability calculations. Radio system according to claim 16, characterized in that the determining means (314) in determining the probability of the value of at least one parameter describing tai. Radio system according to any one of the preceding claims 11-17, characterized in that the determining means (314), in determining 10, calculate the probability of a change in the value of at least one parameter which describes tai and which is included in the speech frame to be processed. Radio system according to claim 18, characterized in that in the decision means (314) a threshold value is determined for the probability of a change in the value of the parameter during a certain number frame number. Radio system according to claim 19, characterized in that the determining means (314) determine that the speech frame does not contain tai which can be decoded with the speech decoder (316), if the probability of a change is less than the threshold value. A subscriber terminal in a radio system comprising a channel decoder 20 (310) for channel decoding a channel-coded speech frame and a speech decoder (316) for decoding the speech frame; Characterized by the fact that the subscriber terminal (112, 114) also comprises determining means: ***; (314) to determine on the basis of the value of at least one parameter which · ···. ^. ·. : 25 writes tai and which is included in the channel-decoded speech frame, whether the speech frame contains * tai which can be decoded with the speech decoder (316), if the speech frame according to the channel decoder (310) is error-free; and the speech decoder (316) is arranged to decode the speech frame if the speech frame according to the decision contains tai which can be decoded with the speech decoder: "30 (316), and the speech decoder (316) is arranged not to decode the speech frame according to the decision which does not contain tai which can is decoded with the speech decoder * j: *. (316), the deciding means (314) being arranged to transmit a return sequence. ·· *. (326) to the speech decoder (316), the speech decoder (316) being arranged to insert a ** * Pause at the location of the speech frame in question, or the deciding means (314) are arranged to cause an interruption in the radio connection. A network of a radio system comprising a channel decoder (310) for channel decoding a channel-encoded speech frame and a speech decoder (316) for decoding the frame; characterized in that the network (110) further comprises determining means (314) for determining on the basis of the value of at least one parameter describing tai and which is included in the channel decoded tair frame, whether the tair frame contains tai which can be decoded by the speech decoder ( 316), if the tier frame according to the channel decoder (310) is error free; And the speech decoder (316) is arranged to decode the tai frame if the tai frame according to the decision contains tai which can be decoded by the speech decoder (316), and the speech decoder (316) is arranged not to decode the tai frame if the tai frame does not contain tai which can be decoded with the speech decoder. (316), wherein the decision means (314) are arranged to send a return sequence (326) to the speech decoder (316), the speech decoder (316) is arranged to insert into the speech a pause at the location of the tary frame in question, or the decision means (314) are arranged to cause a break in the radio connection. · · · · · · · · ···· · · · · · 1 · · · ··· · · · 1 ·· ·· 1 • ·· • · • ♦ · • · • «• 1 · • ♦ 1 * · • · • · 1 # · * · · ·· «» · * · ··· • · • # · * 1 · «· 1« # · • · ··· -. · • l • · · • 1 · • · · *